Environ Geol (2009) 57:1521–1535
DOI 10.1007/s00254-008-1429-3
ORIGINAL ARTICLE
Spatial relationship of groundwater arsenic distribution
with regional topography and water-table fluctuations
in the shallow aquifers in Bangladesh
M. Shamsudduha Æ L. J. Marzen Æ A. Uddin Æ
M.-K. Lee Æ J. A. Saunders
Received: 14 February 2007 / Accepted: 9 June 2008 / Published online: 27 June 2008
Ó Springer-Verlag 2008
Abstract The present study has examined the relationship
of groundwater arsenic (As) levels in alluvial aquifers with
topographic elevation, slope, and groundwater level on a
large basinal-scale using high-resolution (90 m 9 90 m)
Shuttle Radar Topography Mission (SRTM) digital eleva-
tion model and water-table data in Bangladesh. Results
show that high As ([50 lg/l) tubewells are located in low-
lying areas, where mean surface elevation is approximately
10 m. Similarly, high As concentrations are found within
extremely low slopes (\0.7°) in the country. Groundwater
elevation (weekly measured by Bangladesh Water Devel-
opment Board) was mapped using water-table data from
950 shallow (depth \100 m) piezometers distributed over
the entire country. The minimum, maximum and mean
groundwater elevation maps for 2003 were generated using
Universal Kriging interpolation method. High As tubewells
are located mainly in the Ganges–Brahmaputra–Meghna
delta, Sylhet Trough, and recent floodplains, where
groundwater elevation in shallow aquifers is low with a
mean value of 4.5 m above the Public Works Datum (PWD)
level. Extremely low groundwater gradients (0.01–
0.001 m/km) within the GBM delta complex hinder
groundwater flow and cause slow flushing of aquifers. Low
elevation and gentle slope favor accumulation of finer
sediments, As-carrying iron-oxyhydroxide minerals, and
abundant organic matter within floodplains and alluvial
M. Shamsudduha
L. J. Marzen
A. Uddin
M.-K. Lee

J. A. Saunders
Department of Geology and Geography,
Auburn University, Auburn, AL 36849, USA
M. Shamsudduha (&)
Department of Geography, University College London,
Pearson Building, Gower Street, London WC1E 6BT, UK
e-mail: m.shamsudduha@ucl.ac.uk
deposits. At low horizontal hydraulic gradients and under
reducing conditions, As is released in groundwater by
microbial activity, causing widespread contamination in
the low-lying deltaic and floodplain areas, where As is
being recycled with time due to complex biogeochemical
processes.
Keywords Arsenic
SRTM digital elevation model

Spatial distribution
Groundwater gradient
Bangladesh
Introduction
High level of dissolved arsenic (As) concentration in allu-
vial aquifers of Bangladesh and West Bengal, India has
been recognized as the worst groundwater contamination in
the world (Smith et al. 2000; British Geological Survey and
Department of Public Health Engineering, Bangladesh
2001, hereafter referred as BGS and DPHE 2001). Millions
of people are now exposed to the elevated As concentra-
tions in groundwater, of which the majority population
(*120 million) lives in the floodplains and low-lying
Ganges–Brahmaputra–Meghna (GBM) deltaic areas of the
south-central parts of Bangladesh and southern part of West
Bengal, India (Mukherjee et al. 2007). Over the last few
decades, residents have been using millions of hand-oper-
ated tubewells (estimated 10–12 millions) for extracting
groundwater for drinking and approximately 1 million
shallow (\100 m) wells for dry-season irrigation purposes
from alluvial aquifers, of which many contain dangerously
high levels of dissolved As (Smith et al. 2000; BADC 2003;
Harvey et al. 2006). Geochemical data of 3,534 wells
compiled in the National Hydrochemical Survey (NHS) in
Bangladesh were collected by the Department of Public
Health Engineering, Bangladesh and British Geological
123
1522
Survey and Mott MacDonald Ltd. (UK) (BGS and DPHE
2001). Gaus et al. (2003), using the same dataset, have
estimated that nearly 35 million people are drinking
groundwater containing As of a concentration of more than
50 lg/l (Bangladesh standard) and nearly 57 million people
are estimated to be exposed to a concentration exceeding
10 lg/l (World Health Organization standard). It has been
shown in several national- and local-scale studies that dis-
tributions of As in alluvial aquifers in the country are not
uniform (BGS and DPHE 2001; Yu et al. 2003; van Geen
et al. 2003; Shamsudduha 2004; Hoque et al. 2008; Aziz
et al. 2008). National-scale mapping of groundwater As
concentrations shows that high As-contaminated tubewells
are mostly located in the south-central, southeastern, wes-
tern, and northeastern parts of Bangladesh (Fig. 1), whereas
tubewells in the northern, northwestern, north-central, and
eastern parts have lower concentrations of As in the
groundwater (BGS and DPHE 2001; BAMWSP 2002).
High As groundwater occurrences are mostly seen
within the fluvially influenced active and mature deltaic
deposits, alluvial plains and valley deposits, and recent
floodplains along the major rivers in the country (BGS and
DPHE 2001). Many studies evaluated various aspects of As
contamination in the groundwaters of Bangladesh over the
last decade since the first detection of high As occurrence
in groundwater reported in 1990s (Dhar et al. 1997;
Nickson et al. 1998; Chakraborti et al. 1999; Smith et al.
2000; McArthur et al. 2001; Harvey et al. 2002; van Geen
et al. 2003; Zheng et al. 2004; Ahmed et al. 2004; Saunders
et al. 2005; Shamsudduha and Uddin 2007). The general
consensus is that the reductive dissolution of As-bearing
Fe- and/or Mn-oxyhydroxides is the principal mechanism
leading to high As release in groundwaters in the alluvial
aquifers (Nickson et al. 2000; McArthur et al. 2001; Zheng
et al. 2004; Ahmed et al. 2004; Saunders et al. 2005;
Ravenscroft et al. 2005). Although many studies have
been conducted on the geochemical aspects of groundwater
As in aquifers (Nickson et al. 2000; Zheng et al. 2004,
Saunders et al. 2005), very few works focused on the
control of regional-scale surface topography (Sham-
sudduha and Uddin 2007) and local-scale groundwater
level dynamics on spatial As distributions (Ravenscroft
et al. 2005; Harvey et al. 2006; Stute et al. 2007). Low
surface elevation and sluggish groundwater movement
within a narrow range of negligible hydraulic gradients can
facilitate the accumulation of finer sediments enriched with
As-adsorbing iron-oxyhydroxides minerals and organic
matter that may lead to high As concentrations in
groundwater (Harvey et al. 2002; Ravenscroft et al. 2005;
Shamsudduha et al. 2006; Mukherjee et al. 2007). A pre-
liminary study between spatial As distribution and surface
elevation using an interpolated (300 m 9 300 m horizontal
resolution) digital elevation model (DEM) found that the
123
Environ Geol (2009) 57:1521–1535
surface elevation shows a negative relationship with
groundwater As distribution (Shamsudduha 2004). Another
recent study (Shamsudduha and Uddin 2007) has shown
that high As ([100 lg/l) concentrations are observed in the
south-central deltaic areas where the hydraulic gradients
are very small (0.10–0.01 m/km). In this study, a more
reliable and higher horizontal resolution (90 m 9 90 m)
DEM from the NASA Shuttle Radar Topography Mission
(SRTM) is used to examine the spatial relationship
between groundwater As, surface elevation and slope using
Geographic Information System (GIS) techniques. In
addition, spatial relationship between groundwater As
concentrations and water-level fluctuations (yearly maxi-
mum, minimum and mean hydraulic head distributions) of
2003 within the shallow aquifers (\100 m) has been
examined on the national-scale to improve our under-
standing of hydrological constraints on spatial As
distributions in groundwater.
Description of study area
Bangladesh occupies much of the Bengal Basin, which has
been the major depocenter of sedimentary flux from the
Himalayas and Indo-Burman Ranges drained by the Gan-
ges–Brahmaputra–Meghna river system (Goodbred and
Kuehl 2000). The basin is bounded by the Himalayas to the
distant north, the Shillong Plateau to the immediate north,
the Indo-Burman Ranges to the east, the Indian Craton to
the west, and the Bay of Bengal to the south (Fig. 1; Uddin
and Lundberg 1998). The basin includes one of the largest
delta complexes (GBM delta) in the world.
The alluvial plains of the GBM delta slope from north to
south on a regional-scale, but are interrupted locally by
ridges and tectonically developed depressions, such as
Sylhet Trough and Atrai depression. The Bengal Basin is
comprised of lowland floodplain and delta plain, and is
surrounded by the Tertiary hills of various tectonic origins
(Goodbred and Kuehl 2000; Ravenscroft et al. 2005).
Within the Bengal Basin, the Madhupur and Barind tracts,
which are the uplifted alluvial deposits of Pleistocene age,
interrupt the regional surface gradient of the central basin
(Morgan and McIntire 1959).
Surface elevation in the country ranges from less than
1.0 m in the south to about 1000 m in the southeastern
Chittagong Hill Tracts (CHT). The elevation in northwestern
parts (Teesta Alluvial Fan) of Bangladesh is higher than rest
of the geomorphic units except for the CHT. The Himalayan
foothills are extended up to the north of the Teesta Alluvial
Fan. Elevation decreases gradually from the Teesta fan areas
in the northwestern parts of Bangladesh towards the south-
east in the lower GBM delta. The Madhupur and Barind
tracts are also elevated areas (15–40 m) in the basin. The
elevations of the Ganges–Brahmaputra–Meghna delta and
Environ Geol (2009) 57:1521–1535
Fig. 1 Simplified geological
map of the Bengal Basin.
Groundwater wells from the
National Hydrochemical Survey
(BGS and DPHE, 2001) in
Bangladesh with As
concentrations are shown on
this map
recent floodplains are fairly low (\1–10 m). Elevation in
the Sylhet Trough, which is located between the Madhupur
Tract and Eastern Fold Belt, is also low (Fig. 1). To the
immediate north of the Dauki Fault in Sylhet (Fig. 1), the
surface elevation increases abruptly to 500 m or more in the
Shillong Plateau. In the eastern GBM delta, the surface
elevation is less than 10 m, with a minimum of less than 1 m
in the south.
Landforms in the floodplains and deltas are mainly
characterized by natural levees, and crevasse splays, allu-
vial sands, and channel fill deposits. Large marshes and
peat lands characterize the Sylhet Trough, which is fre-
quently flooded during the monsoon (June–September).
Shallow depressions and abundant peat basins are found in
the central part of GBM delta. Numerous tidal creeks and
mangrove forests characterize the southern delta plain.
1523
Materials and methods
Groundwater arsenic dataset
Geochemical data from the National Hydrochemical
Survey (NHS) (BGS and DPHE 2001) in Bangladesh are
used in this study. This public-domain geochemical
database is available online at http://www.bgs.ac.uk/
arsenic/bangladesh/datadownload.htm (BGS and DPHE
2001). The survey analyzed a total of 3,534 tubewells of
different depths in Bangladesh (Fig. 1). A total of 3,043
As-concentration data from the shallow (depth \100 m)
tubewells have been used in this study. About 27% of
these surveyed tubewells contain high groundwater As
([50 lg/l) concentrations (BGS and DPHE 2001). This
dataset has been chosen because of its high reliability in
123
1524
terms of data collection, processing and analysis, and its
country-wide coverage. This database contains location
coordinates (geographic latitude and longitude) for all
sampling sites that were taken by a hand-held Global
Positioning System (GPS) during the survey. Details on
water sampling and analytical procedures can be found in
BGS and DPHE (2001).
SRTM digital elevation dataset
The Shuttle Radar Topography Mission (SRTM) digital
elevation data, produced by NASA originally, is a major
breakthrough in digital mapping of the world, and provides
a major advance in the accessibility of high-quality ele-
vation data for large portions of the tropics (CGIAR-CSI
2008). The NASA SRTM has provided digital elevation
data (DEMs) for over 80% of the globe (USGS 2002). This
high-resolution digital elevation dataset is particularly
useful in mapping topography in the developing countries
like Bangladesh where expensive digital elevation data
acquisition is not always possible. The SRTM was tasked
to collect 3D measurements of the Earth’s surface in 2000.
The objective of the SRTM mission was to obtain elevation
data on a near-global scale and generate the most complete
high-resolution digital topographic database of the Earth
(USGS 2002). Initially, SRTM data (3-arc second, 90-m
resolution) were available to freely download from the
National Map Seamless Server of the USGS public-domain
repository (online at http://seamless.usgs.gov). In the
unfinished SRTM product, there were numerous voids in
space that are mostly associated with mountain valleys,
water bodies, and sand dunes around the world. Although
many voids were filled in the finished product, yet there
was a significant amount of voids still present, which can
cause substantial problem in deriving slopes and ground-
water modeling. The CGIAR-Consortium for Spatial
Information (http://csi.cgiar.org) has filled all the gaps in
the SRTM data using TOPOGRID algorithm with auxiliary
DEM and made available for download as 5° 9 5° tiles, in
the geographic coordinate system with WGS84 datum
(Jarvis et al. 2006; CGIAR-CSI 2008). In this study, a total
of four SRTM tiles (version 3) were downloaded (as
GeoTiff file format) to cover all of Bangladesh (Fig. 2).
The vertical accuracy of the finished 3-arc SRTM images
was within 16 m depending on types of terrain. However,
in the CGIAR-CSI version 3, the SRTM has significantly
improved the vertical accuracy of the DEM to around 10 m
after void filling (Jarvis et al. 2006; Gorokhovich and
Voustianiouk 2006). However, the error in elevation can be
as high as 15 m depending on the presence of dense forest
(e.g., Sundarbans mangrove), where canopy heights can
reach up to 15–20 m from the ground surface. The surface
elevations in the Sundarbans are generally less than a few
123
Environ Geol (2009) 57:1521–1535
meters (WARPO 2000). In this study, the elevation values
within the Sundarbans have been disregarded since there
are no groundwater As data available for that part of the
country in the NHS database (Fig. 1).
Groundwater elevation dataset
Hydraulic head data were collected from the Bangladesh
Water Development Board (BWDB) for the year 2003.
BWDB has a network of about 1,256 water-table monitor-
ing stations throughout the country where the water-table
(hydraulic head below the piezometer top) is measured once
a week. Each well is established with a reference level,
known as Public Works Datum (PWD). This horizontal
PWD was originally thought to be located at the mean sea
level (msl), but there is a difference between these two
reference levels (PWD is approximately 0.45 m below the
msl; see http://www.ffwc.gov.bd/). Water-table data from
950 shallow (well depth \100 m; mean depth *20 m)
piezometers for the year of 2003 have been used for creating
water level elevation maps on the national-scale (Fig. 3).
Reading of the water-table at each piezometer is converted
into groundwater elevation by subtracting from the refer-
ence level (PWD) before GIS analysis and mapping.
Water-table elevation data for normal hydrological year
(2003) were used in this study since there was not any
major and extensive flood event or drought in Bangladesh
(FFWC 2007) that could possibly affect the natural
groundwater levels in the shallow aquifers. However, it has
been reported that high irrigation (deep tubewells and
motor-driven shallow pumps) during the dry water-stressed
period (December–April) can impact the groundwater
levels (Harvey et al. 2006; Klump et al. 2006), although the
actual impacts and their spatial extents have not been fully
investigated. Maximum, minimum and mean water-level
elevations during 2003 were mapped over the entire
country to compare the spatial relationship between
hydraulic head distributions and groundwater As levels.
Spatial GIS and statistical analyses
A database of groundwater As concentrations with XY
coordinates (Universal Transverse Mercator projection in
Zone 46 North) was used to create a vector point map of
the 3,043 shallow tubewells. Four separate SRTM images
were mosaiced after re-projecting into the UTM system
with the above-mentioned zonal attributes to provide a
single DEM of Bangladesh with the original horizontal
resolution of SRTM (Fig. 2). The original image contains
elevation information in each cell as floating-points, in
which the cell values commonly represent continuous
surfaces, assigned as decimal points (USGS 2002). The
SRTM surface elevation data, which are remotely sensed,
Environ Geol (2009) 57:1521–1535 1525

Fig. 2 Digital elevation model

(DEM) of Bangladesh and
groundwater As concentrations
in the National Hydrochemical
Survey tubewells. DEM was
created from SRTM data from
the Seamless Data Distribution
System of USGS. Wells
exceeding As concentrations of
50 lg/l are used for mapping.
High As wells are mainly
located within topographically
low areas. Transects A–B and
C–D show the relationship
between As and elevation in
Fig. 7

are not highly accurate for areas of dense forests and very
flat topography such as Bangladesh. One of the major
problems with the SRTM data for flat areas is the pres-
ence of sinks or depressions in low-lying areas that are
not necessarily associated with small river channels,
shallow swamps and lakes in floodplains. Before using the
SRTM data for any spatial GIS analysis, the depressions
were filled using the fill-sink function under hydrology
tool in the spatial analyst extension of ArcGIS 9.2 pro-
gram. After removing negative values (associated with
river channels) and artificial topographic sinks from the
SRTM data, a more reliable DEM was generated with the
same spatial resolution of 90 m 9 90 m for GIS analysis
(Jarvis et al. 2006). The NHS tubewell data have GPS
coordinates but do not have the surface elevation at their
locations. The tubewell coordinates were used to extract
specific elevation for each sample location by extracting
DEM values (m) to tubewell coordinates by bilinear
interpolation (within a 2 9 2 neighborhood of cells) from
the DEM. Elevation values for each of the 3,043 shallow
tubewell locations were stored in a data file (.dbf) for
statistical analysis.

123
1526 Environ Geol (2009) 57:1521–1535

Fig. 3 Map showing the

locations of 950 shallow (mean
depth *20 m) groundwater
monitoring piezometers from
the Bangladesh Water
Development Board water-table
monitoring network. Water-
table data in 2003 were used for
generating hydraulic head
distribution maps. The locations
of some major district towns are
also shown on this map

Digital elevation model can be used to derive slope and
aspect of any area of interest within the extent of the DEM
raster dataset. In this study, topographic slope (0–90°) was
derived from the DEM for Bangladesh using the surface
analysis tool in the ArcGIS program (Fig. 4). Since a major
portion of the country is covered by low-lying floodplains
and delta plains, the resulting slope is very small except for
the eastern part where the surface topography is rugged due
to folding of the Cenozoic sediments. However, minor
variations in the slope distributions are noticed within the
central and northwestern Bangladesh in the Madhupur and
Barind Tract areas, respectively (Fig. 4). Slope at each
NHS tubewell has been extracted by the same GIS method
described in the previous section.
Groundwater elevation data from the shallow aquifers
were used to create water level surface maps by interpo-
lating from 950 data points using Universal Kriging (UK)
method with an exponential semivariogram model. The
Universal Kriging is one of the most appropriate interpo-
lation methods for contouring groundwater level, which is
generally a non-stationary variable (Gundogdu and Guney
2007). Since the regional groundwater flow in Bangladesh

123
Environ Geol (2009) 57:1521–1535
Fig. 4 Topographic slope
distributions in Bangladesh as
derived from DEM. Tubewells
containing groundwater As
more than 50 lg/l are
concentrated within the GBM
delta, Sylhet Trough and recent
floodplains
is mainly driven by topography (variations in elevation),
there is a general trend (from north to south) in the
groundwater levels. Groundwater level data were detr-
ended using a first-order polynomial regression before
modeling the semivariances and applying for geostatistical
interpolation. Three interpolated groundwater elevation
surface maps (Fig. 5) were generated for three different
hydrological phases during a calendar year: (i) groundwa-
ter elevation map for the maximum water depth during
2003, which corresponds to the lowest water-table that is
generally reached during the driest or highest irrigation
1527
period (March–April); (ii) groundwater elevation map for
the minimum depth during 2003, which corresponds to the
highest water-table that is generally reached during the
peak of monsoon rainfall and flood event (August–Sep-
tember); and (iii) mean groundwater elevation map for
2003, which corresponds to either the onset of monsoon
rainfall or rapidly rising groundwater level (June–July) or
to the natural recession of groundwater level (November–
December). The gridded raster datasets of groundwater
level were clipped within the international boundary of
Bangladesh using a vector dataset of the political boundary
123
1528
(Fig. 5). Interpolated water level elevations at the NHS
tubewell locations were extracted from the mean water
level elevation raster dataset using the extraction function
in spatial analyst extension.
In this study, both descriptive and bivariate statistical
methods were applied to evaluate the relationships between
groundwater As, elevation, slope and groundwater levels.
The descriptive statistical method was applied to summa-
rize the basic statistical parameters of surface elevation,
slope, and groundwater levels for given thresholds (10, 10–
50, and [50 lg/l) of groundwater As concentrations. The
bivariate statistics determine the nature of relationships
between variables and ascertains the strength of their sta-
tistical association. Pearson correlation coefficients were
calculated to evaluate the relationships between ground-
water As, surface elevation, slope and groundwater levels.
Emphasis was given more on the descriptive statistics than
the correlation analysis since the relationships between the
variables are mostly non-linear.
Results and interpretations
Surface elevation and groundwater arsenic
Variations in surface topography and groundwater As
levels are shown in Fig. 2 using DEM and a total of 3,043
shallow tubewells from the NHS database. High As
([50 lg/l) tubewells are located mostly within the low-
lying GBM delta and in Sylhet Trough area (Fig. 2). The
surface elevations of these areas vary from\1 m to as high
as 15 m. Some high As tubewells are located over the
northwestern stable delta parts (south of the Ganges river),
Brahmaputra river floodplains and Atrai depression where
elevation can reach as high as 25–30 m. It should be noted
that the relationship between groundwater As and surface
elevation is non-linear (Fig. 6a). The Pearson correlation
coefficient (r) between As and surface elevation is -0.30
(P \ 0.0001), which suggests that these two variables are
negatively correlated. However, both high ([50 lg/l) and
low (\50 lg/l) groundwater As concentrations are found
within low (\15 m) elevation in the country (Fig. 6a). Two
regional transect lines across the country show the spatial
relationship between groundwater concentrations and var-
iation in surface elevation (Fig. 7). From these N–S and E–
W trending profiles, it is apparent that the relationship
between surface elevation and groundwater As distribution
is negative on regional scales. High groundwater As values
are found within low topographic areas. Within the As hot-
spots, subtle change in surface elevation seems to have
noticeable influence on As concentrations (Fig. 7).
Descriptive statistical analysis (Table 1) shows that 56% of
NHS tubewells with As concentrations of \10 lg/l (in
123
Environ Geol (2009) 57:1521–1535
Fig. 5 Groundwater hydraulic head distribution maps of Bangladesh c
as interpolated from a total of 950 shallow piezometers. Universal
Kriging (UK) method with exponential semivariogram models was
applied for creating these gridded maps. Tubewells with As concen-
trations exceeding 50 lg/l are superimposed on these hydraulic head
maps for visual comparison. The maximum, minimum and mean
hydraulic heads during 2003 are shown here with the general
directions for groundwater flows (arrows on maps) in the shallow
(depth \100 m) aquifer system. High As wells are mostly located in
south-central Bangladesh where hydraulic heads are very low with no
significant variations in the horizontal hydraulic gradients
estimated area of 39,055 km2 excluding the Chittagong
Hill Tracts and the Sundarbans) are located within the
elevated areas with a mean elevation of 22 m (above msl).
Arsenic concentrations of 10–50 lg/l (in estimated area of
35,536 km2) are located in slightly elevated areas where
the mean elevation is approximately 14 m with a standard
deviation of 8.5 m. Areas with the surface elevation of
10 m or less tend to be highly contaminated ([50 lg/l)
with groundwater As, which covers an area of approxi-
mately 54,235 km2 (Table 1). However, the standard
deviation of SRTM elevation within these low-lying areas
is approximately 5 m.
Slope and groundwater arsenic
The spatial relationship between groundwater As distribu-
tions and topographic slope in Bangladesh is shown in
Fig. 4. For more than 90% area of Bangladesh, the topo-
graphic slope is below 4°. Slopes are higher (10–80°) in the
elevated areas in the Chittagong Hill Tracts and Sylhet in
eastern Bangladesh. In the central part, slopes are as high
as 10° only in the Madhupur Tract, which is a Pleistocene
highland. Within the GBM delta, Sylhet Trough, and
alluvial floodplains, the average slope is below 1.0°,
sometimes ranging from 0.1 to 1.5°. In the southern GBM
delta, slopes are slightly higher than the rest of the delta
complex.
Arsenic-contaminated tubewells (NHS dataset) are
located within the low-slope areas, such as GBM delta,
Sylhet Trough and alluvial floodplains in Bangladesh
(Fig. 4). Although the variations in As concentrations with
the variations in slope (all samples, n = 3,040) are not
significant on a regional-scale as suggested by the corre-
lation coefficient (r = - 0.05, P \ 0.10), the correlation
between As concentrations and particularly high slopes
([1.5°; n = 200) is negative (r = - 0.17, P \ 0.05). A
scatterplot between groundwater As and topographic slope
in Fig. 6b shows the association of high As with very low
surface slope in the country. Tubewells with low concen-
trations of As are also found within low slope areas, but
high As wells are not found in the areas of high slope
([1.0°). Table 1 shows the relationship between slope and
groundwater As with other statistical attributes. Wells with
Environ Geol (2009) 57:1521–1535 1529

123
1530 Environ Geol (2009) 57:1521–1535

Fig. 6 Bivariate scatter plots

with best fit lines show
relationship between
groundwater As, SRTM
elevation, slope and
groundwater levels in shallow
aquifers. a shows the non-linear
relationship between
groundwater As and surface
elevation. The overall
relationship is negative. High
As concentrations ([50 lg/l)
are mainly found at low
elevations (\10 m above msl;
see inset graph); b scatterplot
shows the non-linear
relationship between
groundwater As and slope. The
overall relationship is negative.
High As concentrations are
found at low slopes (\1°; see
inset graph); and c scatter plot
shows the non-linear
relationship between
groundwater As and
groundwater levels (m, PWD).
The overall relationship is
similar to that seen between
groundwater As, surface
elevation and slope. High As
concentrations are found at low
groundwater elevations (\10 m
above PWD; see inset graph)

As levels exceeding 50 lg/l are located in areas where
mean slope is approximately 0.60° with a standard devia-
tion of 0.35°. Low concentrations (\10 lg/l) of As are
found in areas of slightly higher slopes where the mean
slope is *0.70°.
Groundwater elevation and arsenic
Three water-level gridded maps (maximum, minimum and
average hydraulic heads) in the shallow aquifers of Ban-
gladesh for the year 2003 have been generated from the
BWDB monitoring piezometers (Fig. 5). Hydraulic heads
are reported as groundwater elevation (m) with respect to the
PWD, which is located approximately at the mean sea level.
Groundwater elevation maps (Fig. 5) show that the water
levels are higher in the northwestern part of Bangladesh,
where surface elevations are also high. Groundwater eleva-
tions are fairly low (\10 m above PWD) in the GBM delta
plains and display broad depressions in the northeastern
Sylhet region of the country. There is a broad circular
(*30 km in diameter) cone of depression surrounding
Dhaka city where groundwater abstraction has been
exponentially increasing since the early 1980s (Hoque et al.
2007). From these national-scale maps, it is seen that the
groundwater flow paths in the shallow aquifers are towards
the general direction of south–southwest. Hydraulic head
distribution maps suggest that groundwater (depth\100 m)
flows toward the deep submarine canyon of the ‘‘swatch of
no ground’’ in the Bay of Bengal (Fig. 5), but this requires
further investigation and substantiation. Other than varia-
tions in seasonal magnitude of groundwater heads, there are
no strong variations in the regional patterns of groundwater
head distributions. Overall, groundwater flow in the country
is mainly topography-driven and varies with precipitation,
interaction with river channels and irrigation abstractions
(BGS and DPHE 2001; Harvey et al. 2006). Around Dhaka
city, groundwater elevation is exceptionally low (less than
-30 m) due to extremely high withdrawal (domestic and
industrial water supplies) of groundwater that exceeds the
natural recharge rate by several orders of magnitude (Hoque
et al. 2007). High As tubewells are located within the areas of
low groundwater levels in Bangladesh (Fig. 5). Relationship
between groundwater As and shallow piezometer hydraulic
heads is non-linear (Fig. 6c) and inverse as suggested by the

123
Environ Geol (2009) 57:1521–1535 1531

Fig. 7 Two regional transects

(a) A–B (north–south) and (b)
C–D (west–east) across
Bangladesh show the variations
in groundwater As
concentrations against surface
elevation (modified from
Shamsudduha and Uddin 2007).
In south-central Bangladesh,
where surface elevation is
very low, most wells are
contaminated with high As
levels

negative Pearson correlation coefficient (r = -0.28, Discussion and conclusions

P \ 0.0001). High As values are mostly found within very
low (\10 m) groundwater level areas, although co-existence
of both high and low As wells is not uncommon. Descriptive
statistical analysis shows (Table 1) that approximately 27%
of NHS tubewells with As concentrations exceeding 50 lg/l
are located within areas where the mean water-table eleva-
tion is approximately 4.5 m with a median of 3.25 m. Low
As (\10 lg/l in estimated 39,055 km2) tubewells are located
in the areas where the mean hydraulic head in 2003 was 16 m
(above PWD) with a median of 11 m. Tubewells with
moderate As concentrations (10–50 lg/l) are found where
the mean hydraulic head is 9.2 m with a median of 7.0 m
above the PWD (Table 1).
In this study, high levels ([50 lg/l) of As are generally
found in areas of low surface elevation, gentle slopes, and
low groundwater levels on a regional-scale in the Bengal
Basin. All these components are inter-related and are
controlled by geology, geomorphology, precipitation and
abstraction (irrigation and urban water supplies). Surface
elevation and topographic slopes, and to a lesser extent,
heterogeneity in fluvial geology, are the main factors that
greatly control regional-scale groundwater flow within the
shallow alluvial aquifers in the basin. Although the nega-
tive correlation coefficients between groundwater As,
surface elevation and slope are not statistically significant,

123
1532 Environ Geol (2009) 57:1521–1535

Table 1 Descriptive statistics on the relationships within surface elevation, topographic slope, groundwater elevation, and arsenic concentra-
tions in Bangladesh

Surface elevation (SRTM, m) and groundwater arsenic

Statistical parameters Elevation at As \ 10 lg/l Elevation at As 10–50 lg/l Elevation at As [ 50 lg/l

Number of As data points 1687 539 813

2
As-affected area cover (in km ) 39,055 (30%) 35,536 (28%) 54,235 (42%)
Range (surface elevation) 0–98 0–77 0–55
Mean elevation (m) 22.2 14 10.6
Median (m) 18 13 10
95% confidence interval 0.68 0.72 0.35
Standard deviation 14.2 8.51 5.16
Variance 203 72 27
Percentage (NHS shallow tubewells) 56% 18% 26%
Slope (degree) and groundwater arsenic
Slope at As \ 10 lg/l Slope at As 10–50 lg/l Slope at As [ 50 lg/l

Number of As data points 1687 539 814

As-affected area cover (in km2) 39,055 (30%) 35,536 (28%) 54,235 (42%)
Range (topographic slope) 0–5.3 0–3.8 0–2.6
Mean slope (degree) 0.70 0.67 0.60
Median (degree) 0.62 0.62 0.61
95% confidence interval 0.022 0.032 0.024
Standard deviation 0.45 0.38 0.35
Variance 0.21 0.15 0.13
Percentage (NHS shallow tubewells) 55% 18% 27%

Groundwater elevation (average in 2003, in meter with respect to PWD) and arsenic
Water elevation at As \ 10 lg/l Water elevation As 10–50 lg/l Water elevation As [ 50 lg/l

Number of As data points 1677 536 811

As-affected area cover (in km2) 39,055 (30%) 35,536 (28%) 54,235 (42%)
Range (groundwater elevation) -21 to 75 -16 to 67 -5 to 50
Mean water level (m) 16 9.2 4.5
Median (m) 10.95 6.9 3.25
95% confidence interval 0.72 0.73 0.31
Standard deviation 15 8.7 4.5
Variance 226 76 21
Percentage (NHS shallow tubewells) 55% 18% 27%

Variations in SRTM elevation, slope and groundwater elevation within three categories of As concentrations (safe As level \10 lg/l; moderate
As level 10–50 lg/l and high As level [50 lg/l) are given above. Areas within these three categories of As concentrations in the country are
estimated from gridded As map produced by spatial GIS method. The Chittagong Hill Tracts and Sundarbans are excluded from the area
estimation

nevertheless, it is clear from the results of both bivariate
correlation and descriptive analyses that high concentra-
tions of As are associated with low elevation and low slope.
Similar relationships have been reported in the Pannonian
Basin of Hungary (Varsanyi and Kovacs 2006), and the
Bassac and Mekong River banks and alluvium deposits in
Cambodia (Buschmann et al. 2007), where most of the high
As wells are located within topographically low areas.
Coexistence of high and low As tubewells within short
spatial distances and even in the same aquifer suggests
greater spatial heterogeneity of sediments in the shallow
aquifers as discussed in Weinman et al. (2006). Their study
also described how high-energy rivers during the Holocene
time in the GBM delta interacted with the low-lying
floodplains and deltaic landforms and created a complex
sedimentological framework, which accounts for the

123
Environ Geol (2009) 57:1521–1535
arsenic heterogeneity over very small spatial scales (i.e.,
within 100 m). Moreover, heterogeneity in sediment size
controls the variations in permeability and hydraulic con-
ductivity in aquifers (Hoque et al. 2008; Aziz et al. 2008).
Generally, hydrologic flushing rates are higher in coarser
sediments (e.g., sand bar), and lower in low-energy
deposits (e.g., overbank floodplain deposits) and these
variable fluxes can cause small-scale variations in
groundwater As concentrations.
Groundwater As distributions are shown to be associated
with the groundwater elevation of shallow aquifers as
depicted from the piezometer water levels. GIS analysis
shows that the distribution of groundwater hydraulic heads
follows the regional surface elevation. High As tubewells
are found within the areas of low (\10 m above PWD)
groundwater elevations. Results suggest a negative
correlation between high As concentrations and low
groundwater levels in the country. Variations in hydraulic
heads are relatively higher in the high abstraction areas
(e.g., Rajshahi district; Fig. 5) between monsoon and
summer periods (Fig. 5). Shallow groundwater heads
fluctuate within 10’s of centimeters to a few meters (up to
10 m with a median of 3.5 m) in most of the country.
Within the As-affected ([50 lg/l) areas (Fig. 5), the
annual fluctuation (based on the range of hydraulic head
distributions in 2003) of shallow groundwater heads is
smaller where the mean head is 3.0 m with a standard
deviation of 1.5 m. Groundwater levels in shallow aquifers
in the GBM delta areas rise almost to the surface level
during the monsoon seasons, although differences between
the dry season and monsoon hydraulic heads in the other
parts of the country are higher than in the GBM areas
(Aggarwal et al. 2000). Shallow aquifers in most of the
country reach the full saturation condition almost every
year with heavy monsoon rainfall and flooding, particularly
in the low-lying areas (Harvey et al. 2006). Groundwater
flow within the shallow aquifers varies from a local- to
intermediate-scale (1,000 m to 10 km) to a regional-scale
(10–100 km), which is mainly driven by variations in
topographic elevation, slope and irrigation abstractions
(Ravenscroft et al. 2005; Mukherjee et al. 2007). The
natural groundwater flow is thought to be greatly disrupted
by large-scale irrigation pumping during the dry period
when vertical flow overtakes the lateral groundwater flow
(Ravenscroft et al. 2005; Harvey et al. 2006). Regional
geomorphic features such as the Tertiary hills, Pleistocene
terraces and major rivers control the groundwater flow-
system within the country (BGS and DPHE 2001).
Groundwater flow is greatly dependent on variations in
the regional hydraulic gradients, except where the natural
groundwater flow is disrupted by dry-season irrigation
pumping (Harvey et al. 2006; Klump et al. 2006). In gen-
eral, hydraulic gradients are small throughout the country
1533
in the absence of dry-season irrigation pumping (Harvey
et al. 2006). Hydraulic gradients in the shallow aquifers
seem to vary both in time and space due to precipitation,
evapotranspiration and variations in aquifer hydrogeology
and hydraulic parameters. BGS and DPHE (2001) reported
that the hydraulic gradients vary in Bangladesh from about
1.0 m/km in the northern part to 0.01 m/km in the south. In
a different study, the hydraulic gradients were reported as
2.0–0.5 m/km in the north, and 0.10–0.01 m/km in the
south of Bangladesh (BAMWSP 2002). Lateral ground-
water flow within these alluvial aquifers is slow within the
Holocene floodplains where the hydraulic gradient is
commonly very low (0.0001) and the Darcy velocity is
about 2 m/year (Burgess et al. 2002; Ravenscroft et al.
2005). From hydraulic head distribution maps (Fig. 5), the
average horizontal hydraulic gradients of shallow aquifers
in different regions in Bangladesh have been estimated as
follows: (i) hydraulic gradient (dh/dl) in the northwestern
part is 0.35–0.25 m/km (northwest to southeast direction);
(ii) 0.05–0.01 m/km (northwest to southeast direction) in
the central part; and (iii) 0.01–0.001 m/km (north to south–
southwest direction) in the southern part. During floods in
the wet season, very little groundwater flow occurs as the
hydraulic gradients become negligible in most low-lying
places (Harvey et al. 2002). In the GBM delta, As-rich
([50 lg/l) groundwaters in shallow (\100 m) aquifers
mostly occur in broad depressions and low-lying flood-
plains of the major rivers, where groundwater elevation
(regardless of seasonal variations) is hardly above 5 m
relative to the PWD.
Low hydraulic gradients in shallow aquifers can greatly
slow down the aquifer-flushing process and obstruct regu-
lar replenishment by new oxic groundwater (Mukherjee
et al. 2007). Gentle slope and low-energy conditions in
surface water systems favor accumulation of finer sedi-
ments and organic matter within the low-lying delta and
floodplains areas. Consequently, As and other dissolved
metals can be adsorbed by iron-oxyhydroxides, which are
flocculated as fine-grained sediments within low-energy
deltaic and floodplain environments (Dowling et al. 2002).
Arsenic is released in groundwater of the shallow aquifers
by the reductive dissolution process aided by iron-reducing
microbes (Islam et al. 2004), which occurs more com-
pletely where abundant organic matter has accumulated
and groundwater flushing is slow. In addition, sluggish
groundwater flow and low topographic gradients can cause
recycling of As-rich groundwater over a long period of
time within the same shallow aquifers (mostly in the GBM
deltaic areas and floodplains), which are already contami-
nated (Zheng et al. 2004; Harvey et al. 2006; Mukherjee
et al. 2007).
The national-scale spatial distributions of groundwater
As are constrained within the shallow aquifers in low-
123
1534
elevation areas with extremely low topographic relief and
negligible hydraulic gradients. Results suggest that spatial
distributions of As in the lowland deltas and floodplains are
controlled by regional topography and slow flushing.
Results from this study along with proper information on
surface and sub-surface geology, and aquifer distributions
can be used to locate areas of possible low groundwater As
in shallow aquifers at the regional scale. However, it would
be interesting to investigate the relationship between
groundwater As, topography and groundwater level
dynamics at the local scale (e.g., 1–10 km) to better
understand the small-scale heterogeneity in groundwater
As distributions and its mobilization in the shallow aquifers
in Bangladesh.
Acknowledgments The authors like to thank British Geological
Survey (BGS) and Department of Public Health Engineering (DPHE),
Bangladesh, for public domain arsenic data. We thank the Seamless
Data Distribution System, Earth Resources Observation and Science
(EROS) by U.S. Geological Survey (USGS) and the CGIAR-Con-
sortium for Spatial Information for distributing void-filled SRTM
data. We acknowledge Bangladesh Water Development Board
(BWDB) for the groundwater elevation data. Thanks are due to Dr.
Alan Fryar for his thoughtful insight and critical review of this work.
This research was partially supported by grants from the National
Science Foundation (NSF 0352936 and 0445250), and support from
AlabamaView.
References
Aggarwal PK, Basu AR, Poreda RJ (2000) Isotope hydrology of
groundwater in Bangladesh: implications for characterization
and mitigation of arsenic in groundwater. International Atomic
Energy Agency, Vienna, TC Project BGD/8/016
Ahmed KM, Bhattacharya P, Hasan MA, Akhter SH, Alam SMM,
Bhuyian MAH, Imam MB, Khan AA, Sracek O (2004) Arsenic
enrichment in groundwater of the alluvial aquifers in Bangla-
desh: an overview. Appl Geochem 19:181–200
Aziz Z, van Geen A, Stute M, Versteeg R, Horneman A, Zheng Y,
Goodbred S, Steckler M, Weinman B, Gavrieli I, Hoque MA,
Shamsudduha M, Ahmed KM (2008) Impact of local recharge
on arsenic concentrations in shallow aquifers inferred from the
electromagnetic conductivity of soils in Araihazar, Bangladesh.
Water Resour Res (in press)
BADC (2003) Survey Report on Irrigation Equipment and Irrigated
Area in Boro/2003 Season. Bangladesh Agricultural Develop-
ment Corporation
BAMWSP (2002) Groundwater Task Force Final Report. Report
prepared for the Government of the People’s Republic of
Bangladesh by Ministry of Local Government, Rural Develop-
ment & Co-operatives, Local Government Division, Dhaka,
Bangladesh
BGS and DPHE (2001) Arsenic Contamination of Groundwater in
Bangladesh, vol 2. Final Report, BGS Technical Report WC/00/19
Burgess WG, Burren M, Perrin J, Ahmed KM (2002) Constraints on
the sustainable development of arsenic-bearing aquifers in
southern Bangladesh. Part 1. A conceptual model of arsenic in
the aquifer. In: Hiscock KM, Rivett MO, Davidson RM (eds)
Sustainable groundwater development. Geological Society
London, Special Publication, vol 193, pp 45–163
123
Environ Geol (2009) 57:1521–1535
Buschmann J, Berg M, Stengel C, Sampson ML (2007) Arsenic and
manganese contamination of drinking water resources in Cam-
bodia: coincidence of risk areas with low relief topography.
Environ Sci Technol 41:2146–2152
CGIAR-CSI (2008) Void-filled seamless SRTM data V1, 2004,
International Centre for Tropical Agriculture (CIAT), available
from the CGIAR-CSI SRTM 90 m Database. http://srtm.csi.
cgiar.org. Access date 5 March 2008
Chakraborti D, Biswas BK, Basu GK, Chowdhury UK, Chowdhury
TR, Lodh D, Chanda CR, Mandal BK, Samanta G, Chakraborti
AK, Rahman MM, Paul K, Roy S, Kabir S, Ahmed B, Das R,
Salim M, Quamruzzaman Q (1999) Possible arsenic contamina-
tion free groundwater source in Bangladesh. J Surf Sci Technol
15:179–187
Dhar RK, Biswas BK, Samanta G, Mandal BK, Chakraborti D, Roy S,
Jafar A, Islam A, Ara G, Kabir S, Khan AW, Ahmed SA, Hadi
SA (1997) Groundwater arsenic calamity in Bangladesh. Curr
Sci 73:48–59
Dowling CB, Poreda RJ, Basu AR, Peters SL (2002) Geochemical
study of arsenic release mechanisms in the Bengal Basin
groundwater. Water Resour Res . doi:10.1029/2001WR000968
FFWC (2007) The flooded area of Bangladesh every year from 1954
to 2005. Flood Forecasting and Warning Center, Bangladesh.
http://www.ffwc.gov.bd/
Gaus I, Kinniburgh DG, Talbot JC, Webster R (2003) Geostatistical
analysis of As concentration in groundwater in Bangladesh using
disjunctive kriging. Environ Geol 44:939–948
Goodbred SL, Kuehl SA (2000) The significance of large sediment
supply, active tectonism, and eustasy on margin sequence
development: late Quaternary stratigraphy and evolution of the
Ganges–Brahmaputra delta. Sediment Geol 133:227–248
Gorokhovich Y, Voustianiouk A (2006) Accuracy assessment of the
processed SRTM-based elevation data by CGIAR using field
data from USA and Thailand and its relation to the terrain
characteristics. Remote Sens Environ 104:409–415
Gundogdu KS, Guney I (2007) Spatial analyses of groundwater levels
using universal kriging. J Earth Syst Sci 116(1):49–55
Harvey CF, Ashfaque KN, Yu W, Badruzzaman ABM, Ali MA,
Oates PM, Michael HA, Neumann RB, Beckie R, Islam S,
Ahmed MF (2006) Groundwater dynamics and arsenic contam-
ination in Bangladesh. Chem Geol 228:112–136
Harvey CF, Swartz CH, Badruzzaman ABM, Keon-Blute N, Yu W,
Ali MA, Jay J, Beckie R, Niedan V, Brabander D, Oates PM,
Ashfaque KN, Islam S, Hemond HF, Ahmed MF (2002) Arsenic
mobility and groundwater extraction in Bangladesh. Science
298:1602–1606
Hoque MA, Hoque MM, Ahmed KM (2007) Declining groundwater
level and aquifer dewatering in Dhaka metropolitan area,
Bangladesh: causes and quantification. Hydrogeol J 15:1523–
1534
Hoque MA, Khan AA, Shamsudduha M, Hossain MS, Islam T,
Chowdhury SH (2008) Near surface lithology and spatial
variation of arsenic in the shallow groundwater: southeastern
Bangladesh. Environ Geol . doi:10.1007/s00254-008-1267-3
Islam FS, Gault AG, Boothman C, Polya DA, Charnock JM,
Chatterjee D, Lloyd JR (2004) Role of metal-reducing bacteria
in arsenic release from Bengal delta sediments. Nature 430:68–
71
Jarvis A, Reuter HI, Nelson A, Guevara E (2006) Hole-filled SRTM
for the globe Version 3, available from the CGIAR-CSI SRTM
90 m Database. http://srtm.csi.cgiar.org.
Klump S, Kipfer R, Cirpka OA, Harvey CF, Brennwald MS,
Ashfaque KN, Badruzzaman ABM, Hug SJ, Imboden DM
(2006) Groundwater dynamics and arsenic mobilization in
Bangladesh assessed using noble gases and tritium. Environ
Sci Technol 40:243–250
Environ Geol (2009) 57:1521–1535
McArthur JM, Ravenscroft P, Safiullah S, Thirlwall MF (2001) Arsenic
in groundwater: testing pollution mechanisms for sedimentary
aquifer in Bangladesh. Water Resour Res 37:109–117
Morgan JP, McIntire WG (1959) Quaternary geology of the Bengal
Basin, East Pakistan and India. Geol Soc Am Bull 70:319–342
Mukherjee A, Fryar AE, Howell PD (2007) Regional hydrostratig-
raphy and groundwater flow modeling in the arsenic-affected
areas of the western Bengal basin, West Bengal, India.
Hydrogeol J . doi:10.1007/s10040-007-0208-7
Nickson R, McArthur J, Burgess W, Ahmed KM, Ravenscroft P,
Rahman M (1998) Arsenic poisoning of Bangladesh groundwa-
ter. Nature 395:338
Nickson RT, McArthur JM, Ravenscroft P, Burgess WG, Ahmed KM
(2000) Mechanism of arsenic release to groundwater, Bangla-
desh and West Bengal. Appl Geochem 15:403–413
Ravenscroft P, Burgess WG, Ahmed KM, Burren M, Perrin J (2005)
Arsenic in groundwater of the Bengal Basin, Bangladesh:
distribution, field relations, and hydrogeological setting. Hydro-
geol J 13:727–751
Saunders JA, Lee M-K, Uddin A, Mohammad S, Wilkin RT,
Fayek M, Korte NE (2005) Natural arsenic contamination of
Holocene alluvial aquifers by linked tectonic, weathering, and
microbial processes. Geochem Geophys Geosyst. doi:10.1029/
2004GC000803
Shamsudduha M (2004) Geostatistical and multivariate statistical
analyses on the widespread arsenic problem in the groundwater
of Bangladesh. MSc, University of Technology Sydney, NSW,
Australia
Shamsudduha M, Marzen LJ, Uddin A, Lee M-K, Saunders JA (2006)
A GIS-based spatial analysis on topographic elevation, slope and
groundwater arsenic in alluvial aquifers of Bangladesh: 3rd
annual GIS Symposium, Auburn University, Alabama, USA
Shamsudduha M, Uddin A (2007) Quaternary shoreline shifting and
hydrogeologic influence on the distribution of groundwater
arsenic in aquifers of the Bengal Basin. J Asian Earth Sci
31:177–194
Smith A, Lingas E, Rahman M (2000) Contamination of drinking-
water by As in Bangladesh. Bull World Health Organ 78:1093–
1103
1535
Stute M, Zheng Y, Schlosser P, Horneman A, Dhar RK, Datta S,
Hoque MA, Seddique AA, Shamsudduha M, Ahmed KM, van
Geen A (2007) Hydrological control of As concentrations in
Bangladesh groundwater. Water Resour Res 43. doi:10.1029/
2005WR004499
Uddin A, Lundberg N (1998) Cenozoic history of the Himalayan-
Bengal system: Sand composition in the Bengal Basin, Bangla-
desh. Geol Soc Am Bull 110:497–511
USGS (2002) Shuttle Radar Topography Mission (SRTM) Elevation
Dataset. Sioux Falls, South Dakota. http://seamless.usgs.gov.
Cited 19 February 2006
van Geen A, Zheng Y, Versteeg R, Stute M, Horneman A, Dhar R,
Steckler M, Gelman A, Small C, Ahsan H, Graziano J, Hussein I,
Ahmed KM (2003) Spatial variability of arsenic in 6000 tube
wells in a 25 km2 area of Bangladesh. Water Resour Res
39(5):1140. doi:10.1029/2002WR001617
Varsanyi I, Kovacs LO (2006) Arsenic, iron and organic matter in
sediments and groundwater in the Pannonian Basin, Hungary.
Appl Geochem 21:949–963
WARPO (2000) National water management plan project, draft
development strategy, main final, vol 2. Water Resources
Planning Organization, Bangladesh
Weinman B, Goodbred S, Zheng Y, Singhvi A, Nagar Y, Aziz Z, van
Geen A (2006) Fluvio-deltaic processes and geomorphic devel-
opment at the scale of 100 to 1000 m: their importance in
governing the heterogeneity of groundwater arsenic in Araiha-
zar, Bangladesh. Geological Society of America Abstracts with
Programs 38(7):180
Yu WH, Harvey CM, Harvey HF (2003) Arsenic in groundwater in
Bangladesh: a geostatistical and epidemiological framework for
evaluating health effects and potential remedies. Water Res Res
39(6):1146. doi:10.1029/2002WR001327
Zheng Y, Stute M, van Geen A, Gavrieli I, Dhar R, Simpson J,
Ahmed KM (2004) Redox control of arsenic mobilization in
Bangladesh groundwater. Appl Geochem 19:201–214
123